Research On Valve Vision Grabbing System Based On SVM Calibration Algorithm

Loading and unloading is a necessary process in the production line.The traditional production line is mainly completed by manpower.Now the industry has developed to the stage of automation and intelligence.In the modern production line,the loading and unloading operation is realized by configuring the industrial robot system.However,it is necessary to select the appropriate sorting and grabbing scheme according to the characteristics of the target workpiece and its placement in the material box.For the valve workpieces which are placed irregularly in the material frame,have small size and have no obvious features of surface circle and end face,the robot can not grasp the workpieces automatically by using simple teaching and programming method.Therefore,in order to realize the automation of valve blank loading and unloading in the process,this paper,referring to a large number of literature and application examples of visual system,develops a grasping scheme for valve workpieces.In order to solve the problem of sorting valves one by one from overlapping and multiple workpieces,starting with the surface characteristics and placement characteristics of valve workpieces,this paper puts forward a scheme of using grating projection stereo vision technology to solve the position and orientation of the target workpieces on the basis of target recognition and positioning of the image,guiding the robot to grab the workpieces with high precision and efficiency,and designing the vision guided robot to grab the workpieces The overall structure of the system has certain engineering practical significance and value.The main contents of this paper are as follows:Identify and locate the target workpiece in the image.T he geometric characteristics of the valve workpiece in the image are analyzed,mainly showing the elliptical shape.Through the gray-scale and binary processing of the image,the improved watershed algorithm is used to segment the blob area of the target.Based on the Hough transform ellipse fitting algorithm to extract the ROI area,the target location in the two-dimensional image is completed.Calibration of stereo vision system.Through the analysis of the camera—DLP projector vision system,the system is modeled,including image coordinate system,camera coordinate system,world coordinate system,projector coordinate system,tool coordinate system,etc.In order to solve the conversion relationship between the coordinate systems,the camera calibration is carried out with reference to Zhang's calibration method.The SVM regression algorithm is mainly used to calibrate the grating projection stereo vision model and solve the position and orientation of the target workpiece in space.The eye hand model of eye in eye in eye is calibrated,and the transformation matrix between camera and tool coordinate system is solved,so as to determine the space pose of robot executive end.Test the vision guided robot grabbing system.Solve the mixed programming problem of MATLAB and C + +,establish the communication between the vision system end and ABB Robot end,realize the transmission of the position and attitude data of the execution end,guide the robot movement in the robot studio simulation,and test the rationality of the path planning of the robot execution end in the actual scene.Build a practical vision guided robot grabbing system,and test the grabbing valve workpiece to verify the feasibility of the scheme,the reliability and accuracy of the system.